Abstract
This study presents a new approach to asphalt material analysis, utilising deep generative learning techniques to generate asphalt microstructures from a limited set of micro-CT images. Recognising the critical influence of porosity variation and vertical distribution on the performance and durability of pavement structures, our research focuses on reconstructing the porous microstructure of asphalt. This deep learning-based method enables the rapid creation of new three-dimensional (3D) volumes with controllable porosity profiles, which can be used for material design, simulations and characterisation.
We conduct a comparative analysis between the generated volumes and the original micro-CT images. This approach aims to validate the generated dataset’s accuracy and assess the effectiveness of our method. Our findings indicate that generative deep learning can rapidly generate new representative 3D microstructures with properties that match the real dataset. This advancement provides a new framework for future explorations into developing more durable and efficient pavement systems.
Through the integration of generative deep learning methods with micro-CT imaging techniques, this paper introduces a fresh perspective on the design and analysis of asphalt materials. The developed framework can also be extended to other types of microstructures and properties which can offer a powerful and flexible method in the field of material science and civil engineering.
We conduct a comparative analysis between the generated volumes and the original micro-CT images. This approach aims to validate the generated dataset’s accuracy and assess the effectiveness of our method. Our findings indicate that generative deep learning can rapidly generate new representative 3D microstructures with properties that match the real dataset. This advancement provides a new framework for future explorations into developing more durable and efficient pavement systems.
Through the integration of generative deep learning methods with micro-CT imaging techniques, this paper introduces a fresh perspective on the design and analysis of asphalt materials. The developed framework can also be extended to other types of microstructures and properties which can offer a powerful and flexible method in the field of material science and civil engineering.
